Quick answer: A well-run 2026 podcast agency produces 40+ client episodes per week with a team of six by standardising six steps (brief, record, transcribe, AI edit draft, human QA, publish), running each stage in parallel, and using AI to eliminate the repetitive 60% of post-production. GeraCast or Descript runs the AI layer; Notion or Linear runs the pipeline.

Why this is different from a solo workflow

A solo podcaster optimises for quality on a single show. An agency optimises for throughput across many shows with varying client standards. The problems are different: dependency management, context-switching cost, consistent delivery, and QA at scale. The tools are similar, but the process discipline is stricter.

The six-stage production pipeline

1. Brief & schedule (Monday)

Producer pulls the week's roster from the client tracker. For each episode, a briefing doc is created with: guest bio, three-sentence angle, three primary questions, three follow-ups, client-required mentions, and desired run-time. This is shared with the host 24h before recording.

2. Remote record (Tuesday/Wednesday)

Every recording uses Riverside or GeraCast remote recording for separate multi-track per participant. Guests are sent a pre-record checklist (headphones on, quiet room, backup phone recording running). Live producer joins as an invisible host to monitor levels and cover dropouts.

3. AI first-pass edit (Wednesday/Thursday)

Raw tracks are uploaded to GeraCast, which produces: full transcript, speaker-separated audio, AI-generated first-cut removing silences / filler words / double-takes, chapter markers, show notes draft, and social clips (30-90s auto-selected by engagement prediction). Human time saved: 60-75% vs manual edit.

4. Human QA and polish (Thursday)

Senior editor reviews the AI draft: checks flow, corrects any mis-spoken names, fine-tunes intro/outro, confirms music bed levels, and clears legal (claims, guest corrections). The Descript or GeraCast text-based editor makes this pass a 30-60 minute job for a 45-minute episode.

5. Publish (Friday morning)

One-click distribution to client's host + auto-push to Spotify, Apple, Google, YouTube (as podcast). Show notes publish to the client website via CMS API. Social clips queued in Buffer/Hypefury for timed posting Friday-Monday.

6. Reporting (Friday afternoon)

Weekly client email with per-episode downloads, engagement, top chapters, listener drop-off points, and social-post performance. Automated dashboard link.

Team shape for 40+ weekly episodes

Producers (2): own 8-12 client shows each. Run briefing, scheduling, and client comms.
Senior editors (2): QA the AI-edited drafts. Each handles 10-14 episodes per week.
Publishing coordinator (1): handles distribution, CMS uploads, analytics reporting.
Agency lead (1): sales, onboarding, client escalations, pipeline oversight.

With this team, billable rates of £1,500-£3,000 per episode give agency gross margins of 55-70% once tooling costs are covered (~£500-£1,200/mo total for GeraCast + scheduling + CMS).

Tooling stack

Record: Riverside.fm or GeraCast remote record.
Edit + AI pipeline: GeraCast (AI transcript + first-pass edit + show notes) or Descript.
Host & RSS: Transistor (multi-show on one account) or GeraCast for unified AI+hosting.
Client pipeline: Notion or Linear with per-episode tickets.
Scheduling: SavvyCal team account.
Social distribution: Buffer or Hypefury for clips.
Reporting: Geckoboard or Notion dashboards pulling from GeraCast API.
File storage: Google Drive or Dropbox team.

The operational risks

Guest no-shows / reschedules throw the Tuesday/Wednesday slots. Book a bench list of backup guests.
Client feedback loops are the biggest productivity killer. Enforce one 30-minute review window per episode with clear change categories.
AI hallucination in show notes. Human QA checks: did the AI summarise an opinion as fact; did it misattribute a claim to the wrong speaker?
Platform API changes. Keep the publish pipeline modular so a breaking change in one platform does not stall the whole pipeline.

How a Podcast Agency Produces 40+ Episodes per Week (Workflow Deep-Dive)